Embed Size (px)
Citation preview
Resource
Quantitative Analysis of Fission YeastTranscriptomes and Proteomesin Proliferating and Quiescent CellsSamuel Marguerat,1 Alexander Schmidt,2 Sandra Codlin,1 Wei Chen,3 Ruedi Aebersold,4,5,* and Jurg Bahler1,*1University College London, Department of Genetics, Evolution and Environment and UCL Cancer Institute, London WC1E 6BT, UK2Proteomics Core Facility, Biozentrum, University of Basel, CH-4056 Basel, Switzerland3The Berlin Institute for Medical Systems Biology, Max-Delbruck-Centrum fur Molekulare Medizin (MDC), 13092 Berlin, Germany4Department of Biology, Institute of Molecular Systems Biology, ETH Zurich, CH-8093 Zurich, Switzerland5Faculty of Science, University of Zurich, CH-8057 Zurich, Switzerland
*Correspondence: [email protected] (R.A.), [email protected] (J.B.)
http://dx.doi.org/10.1016/j.cell.2012.09.019
SUMMARY
Data on absolute molecule numbers will empowerthe modeling, understanding, and comparison ofcellular functions and biological systems. We quanti-fied transcriptomes and proteomes in fission yeastduring cellular proliferation and quiescence. Thisrich resource provides the first comprehensive refer-ence for all RNA andmost protein concentrations in aeukaryote under two key physiological conditions.The integrated data set supports quantitative biologyand affords unique insights into cell regulation.Although mRNAs are typically expressed in a narrowrange above 1 copy/cell, most long, noncodingRNAs, except for a distinct subset, are tightlyrepressed below 1 copy/cell. Cell-cycle-regulatedtranscription tunes mRNA numbers to phase-specific requirements but can also bring aboutmore switch-like expression. Proteins greatly exceedmRNAs in abundance and dynamic range, andconcentrations are regulated to functional demands.Upon transition to quiescence, the proteomechanges substantially, but, in stark contrast tomRNAs, proteins do not uniformly decrease butscale with cell volume.
INTRODUCTION
Gene regulation is crucial to implement genomic information
and to shape properties of cells and organisms. Transcriptomes
and proteomes are dynamically tuned to the requirements of
cell volume, physiology and external factors. Although tran-
scriptomic and proteomic approaches have provided ample
data on relative expression changes between different condi-
tions, little is known about actual numbers of RNAs and proteins
within cells and how gene regulation affects these numbers.
More generally, most data in biology are qualitative or relatively
quantitative, but ultimately many biological processes will only
be understood if investigated with absolute quantitative data
to support mathematical modeling. Other areas of science
have long appreciated the limits of relative, or compositional,
data and potential pitfalls of their naive analysis (Lovell et al.,
2011).
Insights into numbers and cell-to-cell variability of selected
mRNAs and proteins have been provided by single-cell studies
(Larson et al., 2009), but these approaches require genetic
manipulation and are not well suited for genome-scale anal-
yses. Relating mRNA to protein abundance in single cells
is challenging, with only one such study available for a
prokaryote (Taniguchi et al., 2010). Global mRNA abundance
for yeast populations have been estimated (Holstege et al.,
1998; Miura et al., 2008). There are no comparisons for cellular
concentrations of mRNAs and the emerging diversity of non-
coding RNAs.
RNA-seq now allows actual counting of RNA numbers,
offering unbiased genome-wide information on average cellular
RNA concentrations in cell populations (Ozsolak and Milos,
2011). Moreover, the global quantification of proteins has
recently become possible owing to advances in mass spectrom-
etry, giving valuable insight into the protein content of different
cells (Beck et al., 2011; Cox and Mann, 2011; Maier et al.,
2011; Nagaraj et al., 2011; Vogel and Marcotte, 2012).
Here, we combine quantitative RNA-seq and mass spec-
trometry to analyze at unprecedented detail and scale how
changes in cell physiology and volume are reflected in the
cellular concentrations of all coding and noncoding RNAs and
most proteins. We analyze two fundamental physiological
states in fission yeast: (1) proliferating cells that need to con-
stantly replenish their RNAs and proteins, and (2) postmitotic
cells that do not grow or divide owing to nitrogen limitation
and reversibly arrest in a quiescent state (Yanagida, 2009).
Although quiescent states are common, both for yeast and
for cells in the human body, most research has focused on
proliferating cells. The ability to alternate between proliferation
and quiescence is central to tissue homeostasis and renewal,
pathophysiology, and the response to life-threatening chal-
lenges (Coller, 2011). For example, quiescent lymphocytes
Cell 151, 671–683, October 26, 2012 ª2012 Elsevier Inc. 671
Figure 1. Transcriptome Quantification in
Proliferating Cells
(A) Abundance distribution of total RNA (green)
and mRNA (black). Red vertical lines indicate 1
and 10 RNA copies/cell, and red hatched lines
delimit expression zones 1 to 3. See also Figure S1
and Table S10.
(B) Abundance for all detected mRNAs (each dot
represents a gene). Green and gray dots corre-
spond to essential and non essential genes,
respectively. Expression zones are indicated at
right.
and dermal fibroblasts become activated to mount immune
responses or support wound healing, respectively. Adult stem
cells also alternate between proliferating and quiescent states,
and the deregulation of either state can cause complex pathol-
ogies such as cancer (Li and Clevers, 2010).
Our integrated transcriptomic and proteomic data, acquired in
parallel under highly controlled conditions in a simple model,
afford varied biological insights and reveal key principles of
RNA and protein expression in proliferating and quiescent cells
with broad relevance for other eukaryotes. This rich resource
also provides a quantitative framework toward a systems-level
understanding of genome regulation, and the common units of
the absolute data allow direct comparison of different biological
processes and organisms.
RESULTS AND DISCUSSION
Transcriptome and Proteome Quantification in TwoConditionsWe acquired quantitative expression data relative to absolutely
calibrated standards for transcriptomes and proteomes of
haploid fission yeast cells. For transcripts, genome-wide mea-
surements were obtained by calibrating RNA-seq data from
total RNA preparations with data on absolute cellular concen-
trations for 49 mRNAs, covering the dynamic expression range.
The overall measurement error was estimated to be �2-fold or
less (Figure S1; Tables S1–S4 available online). Protein quantifi-
cation was performed on the same cell samples using a mass
spectrometry (MS) approach (Schmidt et al., 2011). Selected
proteotypic peptides from 39 proteins (Table S5), covering the
dynamic expression range, were used to absolutely quantify
the corresponding proteins (Tables S6 and S7). These data
were then used to translate the MS-intensities for the other
proteins into estimates of cellular concentration (Figures S2A–
S2D and S3; and Tables S8 and S9). The mean overall measure-
ment error was estimated at 2.4- and 2.7-fold for proliferating
and quiescent cells, respectively.
We quantified transcriptomes and proteomes in two distinct
physiological conditions: (1) exponentially proliferating cells in
672 Cell 151, 671–683, October 26, 2012 ª2012 Elsevier Inc.
defined minimal medium, and (2) quies-
cent cells, 24 hr after nitrogen removal
(Figure S4). We first report the results
from proliferating cells, and then relate
our findings to corresponding data from
quiescent cells. Table S4 provides the cellular copy numbers
for RNAs and proteins in the two conditions.
Most mRNAs Are Expressed in Narrow Range above 1Copy/CellIn proliferating cells, we measured a total of �41,000 mRNA
molecules/cell on average, representing �5% of the overall
�802,000 rRNAs/cell in our samples. Protein-coding genes
produced a median of 2.4 mRNA copies/cell, ranging from
�0.01 to >810 copies (Figure 1A). Only 71 genes showed no
detectable mRNA signal, 43 of which are annotated as
‘‘dubious’’ or ‘‘orphan’’ (Wood et al., 2012). To discuss our
findings, we distinguished three somewhat arbitrary expression
zones, set relative to the one RNA copy/cell mark (Figure 1A).
Zone 1 contained low-abundance mRNAs detected at <0.5
copies/cell. Zone 2 mRNAs were expressed at �1 copy/cell
(0.5–2 copies), where fluctuations due to cell division or
stochastic expression will strongly affect the presence of
mRNAs in cells. Zone 3 mRNAs showedmore robust expression
at >2 copies/cell. Most mRNAs were expressed within a low
and narrow range: whereas >90% of all annotated mRNAs
(4,608/5,110) belonged to zones 2 or 3, 86.1% of these mRNAs
were present at <10 copies/cell (Figure 1A). Low overall
mRNA concentrations have also been reported for budding
yeast, which has comparable gene numbers and cell size, with
even lower estimates for median mRNA abundance (<1 copy/
cell) and total mRNA molecules/cell (Holstege et al., 1998; Miura
et al., 2008). Our findings are in line with a single-cell study of
budding yeast, where five mRNAs show 2.6–13.4 copies/cell,
with a total estimate of 60,000 mRNA molecules/cell (Zenklusen
et al., 2008).
We examined the mRNAs of the 1,273 genes essential for
growth (Kim et al., 2010), which are expected to be expressed
in proliferating cells. Nearly all essential mRNAs were expressed
in zones 2 or 3 (98.4%; Figure 1B). This finding raises the
possibility that �1 mRNA copy/cell defines a natural minimal
threshold for productive gene expression.
The view of �1 mRNA copy/cell as an expression threshold
is supported by recent data from metazoa, where mRNA levels
mRNA metabolic processnucleoluspeptidase activitytransporttransporter activitygrowth module (Chen et al)cytoplasmic translationcytosolic small ribosomal subunitcytosolic large ribosomal subunittranslationribosome biogenesisprotein foldingS phase induced (Rustici et al)generation of precursor metabolites and energycofactor metabolic processligase activityoxidoreductase activitycellular amino acid metabolic processearly meiotic genes (Mata et al)DNA recombinationmiddle meiotic genes (Mata et al)late meiotic genes (Mata et al)shortest mRNAs (Baehler lab)meiosischromosome segregationnitrogen removal induced (Mata et al)stress module (Chen et al)
< 0.5 0.5-2 > 2
1 2 3
mRNA copies/cell=1 <1 <0.1 <0.05 <0.001
pFisher
1 2 3 1 2 3
1 2 3 1 2 3
A B C
D E
Figure 2. Functional Categories and Expression Zones
(A) Hierarchical clustering of p values (Fisher exact test, color-coded as indicated) assessing significance of overlap between genes in functional categories (rows)
and 200-gene sliding windows of mRNA abundance (columns). Vertical red lines delimit the expression zones. Functional categories with p values <0.01 in R1
window are shown. See also Table S11.
(B) Frequency of genes for which corresponding protein is detected in 200-gene sliding window of mRNA abundance (black curve; left axis), together with p
values (Fisher exact test) for significance of overlap between gene list and window (green curve; right axis).
(C) As in (B) for early meiotic differentiation genes (Mata et al., 2002).
(D) As in (B) for core environmental stress response genes (Chen et al., 2003).
(E) As in (B) for ‘‘protein folding’’ genes (Gene Ontology ID: 0006457).
show a bimodal distribution (Hebenstreit et al., 2011): one
group of putative nonfunctional mRNAs present at <1 copy/
cell, and another group of actively transcribed mRNAs ex-
pressed at >1 copy/cell. mRNA levels did not show such
a bimodal distribution in fission yeast (Figure 1A). This disparity
highlights that in differentiated metazoan cells many genes are
not expressed, whereas in proliferating yeast cells most genes
are actively expressed. Notably, when including long noncod-
ing RNAs, which were mostly present at low abundance, fission
yeast also showed a bimodal distribution for transcript levels
(Figure 1A).
Characteristics of Three mRNA Expression ZonesEach expression zone was enriched for distinct functional
categories (Figure 2A), revealing that genes participating in
similar processes typically coordinate their cellular mRNA
concentrations. The mRNA expression zones reflect protein
expression as the 3,397 proteins detected in proliferating cells
showed a strong bias toward highly expressed mRNAs (Fig-
ure 2B), although proteins of low abundance were also confi-
dently detected (see below).
Only 431 genes were present in zone 1 (8.4% of 5,110 protein-
coding genes), which were enriched for meiotic differentiation
functions such as recombination and sporulation (Figures 2A
and 2C). Genes induced during meiosis are tightly repressed
during proliferation, and their expression is regulated at multiple
levels including chromatin (Zofall et al., 2012), transcription (Mata
et al., 2007), and mRNA turnover (Harigaya and Yamamoto,
2007; McPheeters et al., 2009). Only 31 (7.2%) of these genes
produced detectable proteins, most of which were stress
response genes and present near the upper limit of zone 1
(0.5 copies/cell). These findings support the notion that zone 1
genes are not actively transcribed and typically do not lead to
productive protein expression. We propose that the presence
of mRNAs at well below 1 copy/cell reflects active repression
of the corresponding genes. Such low mRNA copy numbers
could be the result of rare stochastic transcription (Hebenstreit
et al., 2011).
The 1,664 genes of zone 2 included 27.8% of all essential
genes, and 880 (52.9%) of these genes produced detectable
proteins. These findings indicate that low mRNA concen-
trations (�1 copy/cell) are compatible with productive gene
Cell 151, 671–683, October 26, 2012 ª2012 Elsevier Inc. 673
Figure 3. mRNA Copy Number Changes during Cell Cycle
Peak (blue) and basal (green) mRNA abundance of cell-cycle-regulated genes
extrapolated from average data in asynchronous cultures, with 10% of cell-
cycle assumed as duration for peak expression. Data for six cell-cycle time
course experiments are indicated by clustered dots (Rustici et al., 2004). Left:
ten histone mRNAs peaking during S phase; right: mik1, mde6, and mei2
mRNAs peaking during M and G1 phases.
See also Figure S5 and Table S12.
expression. Zone 2 genes were functionally enriched for chro-
mosome segregation, nitrogen starvation, and core environ-
mental stress response (Figures 2A and 2D). The latter genes
are rapidly induced in multiple stresses (Chen et al., 2003) and
show highly variable expression across different experimental
conditions (Pancaldi et al., 2010). This enrichment suggests
that �1 mRNA copy/cell corresponds to the basal expression
typical of many stress response genes (Chen et al., 2003). Unlike
the tight repression of meiotic genes, the basal expression of
stress genes could enable a rapid response to sudden environ-
mental challenges. Zone 2 was transitional between zones 1
and 3 also with respect to protein detection (Figure 2B). We
propose that low basal mRNA expression might not always
lead to robust protein expression but might maintain a respon-
sive chromatin environment, e.g., for genes that require rapid
upregulation during stress. Moreover, such low average expres-
sion could reflect a ‘‘bet-hedging’’ strategy to diversify cellular
phenotypes and promote population survival to unexpected
environmental challenges (Lopez-Maury et al., 2008).
Zone 3 contained 2,944 genes (57.6% of all genes), which
were enriched for several functional categories (Figure 2A). For
example, genes involved in translation and protein folding
674 Cell 151, 671–683, October 26, 2012 ª2012 Elsevier Inc.
tended to be highly expressed (Figure 2E). Proteins were de-
tected for 2,486 (84.4%) of the zone 3 genes, indicating that
robust mRNA expression typically results in robust protein
expression.
Together, these data show that mRNAs of different functional
categories are typically expressed in distinct abundance ranges.
The data further support the notion that an expression of �1
mRNA copy/cell defines a minimal threshold for productive
gene expression. We conclude that the three mRNA expression
zones reflect characteristic gene groups with respect to regu-
lation, cellular functions, and protein production.
Effect of Cell-Cycle-Regulated Gene Expressionon mRNA NumbersGlobal studies have revealed hundreds of fission yeast genes
that are periodically expressed during the cell cycle (Marguerat
et al., 2006). The corresponding mRNA copy numbers will
therefore fluctuate, and our quantitative data from asynchronous
cell cultures reflect time-averaged mRNA counts. The effects
of cell-cycle-regulated gene expression on absolute mRNA
abundance are not known. Two scenarios are plausible: periodic
gene expression might boost mRNA numbers for proteins
required at higher levels during certain cell-cycle phases, or it
might act in a switch-like manner to tightly restrict expression
to a specific phase.
To distinguish between these two hypotheses, we applied
simple modeling to extrapolate absolute changes in mRNA
abundance of cell-cycle-regulated genes from our data in asyn-
chronous cultures. The model assumes that periodic genes
peak in expression during a defined cell-cycle phase and
show basal expression during the other phases. We derived
phase-specific mRNA copy numbers for 241 periodic genes
with expression peaking in M, G1, or S phase (Figure 3).
Most of these genes (96.3%) showed variations in mRNA
expression that remained within zones 2 and 3 throughout
the cell cycle. For example, the mRNAs for 10 histone genes
were abundant throughout the cell cycle, with their numbers
peaking during DNA replication (Figure 3). This pattern is
consistent with the idea that periodic gene expression boosts
mRNA numbers to accommodate an increased demand for
histones during S phase, with a high basal requirement in
other phases.
Only nine genes showed a more switch-like pattern of tran-
scription: they belonged to zones 2 or 3 during peak expres-
sion, but dropped to zone 1 during basal expression, thus
crossing the �1 mRNA copy/cell threshold (Figure 3). We
propose that expression of these genes is restricted to a
specific cell-cycle phase, and repressed when they may be
harmful. For example, the mik1 gene encodes an inhibitor of
mitosis with a tightly restricted expression window at both
mRNA and protein levels (Ng et al., 2001). Another example
was mei2, encoding a protein that promotes untimely meiosis
when activated at the wrong time (Harigaya and Yamamoto,
2007). We conclude that periodic gene expression generally
tunes mRNA numbers to specific requirements in different
cell-cycle phases but also, in special cases, reflects regulatory
switches restricting the expression of critical regulators to
specific phases.
Figure 4. Quantitative Analysis of Long Noncoding RNAs
(A) Absolute abundance of mRNAs (gray), and all (dark green), intergenic (bright green), and antisense (blue) lncRNAs. Expression zones are indicated at right.
(B) Cumulative plot of copy numbers contributed by lncRNA genes ranked by decreasing abundance, with genes expressed in zones 2 and 3 at left of red line.
(C) Sequence scores for lncRNAs in libraries made from total versus poly(A)+ RNA. Bright green circles, lncRNAs expressed in zone 1; dark green and orange
triangles, lncRNAs expressed in zones 2 or 3; orange, lncRNAs that are R4-fold more abundant in total than in poly(A)+ RNA library.
See also Table S13.
Long Noncoding RNAs Are Typically Present below 1Copy/CellSubstantial transcriptional activity occurs outside of protein-
coding genes and produces distinct noncoding RNAs. Besides
the well known RNAs involved in gene expression such as
rRNAs, tRNAs, snRNAs, and snoRNAs (Figure S1), 1,557 long
noncoding RNAs (lncRNAs) have been identified in fission yeast
(Rhind et al., 2011; Wilhelm et al., 2008). These lncRNAs are
reminiscent of lincRNAs in multicellular eukaryotes and unanno-
tated transcripts in budding yeast (SUTs, CUTs, XUTs) (Atkinson
et al., 2012), but differ from the short RNAs produced by RNA
interference pathways (Grewal, 2010). In proliferating cells, we
could quantify 86.4% (1,346/1,557) of these lncRNAs, which
together accounted for only 1,672 RNA molecules/cell (Table
S10). Accordingly, 1,159 (85.5%) of these lncRNAs belonged
to zone 1, numbering well below 1 copy/cell, similar to tightly
repressed mRNAs (Figures 1A and 4A). lncRNAs transcribed
both in intergenic regions and antisense to coding genes typi-
cally belonged to zone 1 (Figure 4A). By analogy with meiotic
genes, such low abundance could reflect tight repression at
transcriptional, posttranscriptional, and/or chromatin levels.
The remaining 187 lncRNAs (14.5%) were expressed in zones
2 and 3, at �1–200 copies/cell. Notably, this small group
accounted for >90% of the total cellular number of lncRNA
molecules (Figure 4B). This group was not enriched for lncRNAs
conserved in other fission yeast species (Rhind et al., 2011). The
coding genes that were associated with antisense RNAs ex-
pressed in zones 2 or 3 were more likely to be repressed in
zone 1 (pbinomial < 10�8), consistent with a role of antisense tran-
scription in repressing the corresponding sense transcription.
We compared the sequence scores from RNA-seq libraries
produced from either total or poly(A)-enriched RNA. Most
lncRNAs were present at similar levels in the two libraries, irre-
spective of their abundance (Figure 4C). This result suggests
that most lncRNAs are polyadenylated and therefore likely
transcribed by RNA polymerase II (Pol II). Intriguingly, 38
lncRNAs were much more abundant in the total RNA library
(Figure 4C). These lncRNAs were depleted during poly(A) enrich-
ment and hence likely not polyadenylated. This finding raises
the possibility that these lncRNAs are not transcribed by Pol II,
or that they are matured via poly(A) trimming (Lemay et al.,
2010). These lncRNAs showed no particular sequence features
using Rfam (Gardner et al., 2011), and they were not similar to
any well-known RNAs such as snRNAs or snoRNAs. Remark-
ably, although these lncRNAs made up only 2.4% of the known
lncRNA repertoire, they accounted for 63.6% of the total lncRNA
molecules. Taken together, this analysis uncovered two distinct
classes of lncRNAs that differ based on their absolute expres-
sion and polyadenylation status, with a small class of nonpolya-
denylated lncRNAs contributing the majority of all cellular
lncRNA molecules.
Proteins Greatly Exceed mRNAs in Abundanceand Dynamic RangeIn proliferating cells, we could quantify 3,397 (66.5%) of the
5,110 predicted proteins, adding up to an average of 60.3 million
protein molecules/cell. The identified proteins showed no strong
bias against any protein class (Figure S2E), underlining the
broad coverage achieved. Protein-coding genes produced a
median of 3,919 protein copies/cell, with a dynamic range of
five orders of magnitude (Figure 5A). The most abundant protein
was the translation factor EF-1a (Tef102), expressed at �1.1
million copies/cell, whereas the lowest detectable protein was
the formin Cdc12, expressed at <100 copies/cell. Our data
were similar to quantitative microscopy data for 27 cytokinesis
proteins (Figure S2H) (Wu and Pollard, 2005). On average,
proteins were�1,850 timesmore abundant than their respective
mRNAs. This finding indicates that translation serves as a global
amplification step, although some of this difference could also
reflect longer half-lives for proteins than for mRNAs.
The mRNAs coding for the 3,397 detected proteins were
greatly enriched in expression zones 2 and 3 (Figure 5B). More-
over, the 1,273 essential genes (Kim et al., 2010) produced
a significantly higher proportion of detectable proteins (81.9%,
Cell 151, 671–683, October 26, 2012 ª2012 Elsevier Inc. 675
Figure 5. Quantitative Analysis of Proteome in Proliferating Cells
(A) Abundance distribution for mRNAs (green) and proteins (red). Red vertical lines delimit expression zone 2 (0.5–2 mRNA copies/cell). See also Figures S2, S3,
and Table S10.
(B) Absolute abundance for all mRNAs (each dot represents a gene). Dark and light blue dots correspond to genes for which proteins were detected or not,
respectively.
(C) Protein versus mRNA abundance. Black curve, sliding median.
(D) Protein/mRNA ratio versus protein abundance. Red dots: ribosomal proteins; black curve: sliding median.
(E) Protein abundance for selected functional categories. Each dot represents a protein. Haploid Schizosaccharomyces pombe cells contain 5,110 and 10,220
annotated protein-coding genes in G1 and G2 phase, respectively (red zone), and 5,348 introns across 2,523 intron-containing genes (red and yellow zones,
respectively). In proliferating cells, we measured �41,000 mRNA molecules (dark green line) and 1.1–2.6 3 105 copies of each rRNA (green zone). Ribosomal
proteins copies/cell for paralogs were summed up.
See also Figure S6.
pbinomial < 10�15). The 458 robustly expressed zone 3mRNAs not
associated with detectable proteins were enriched for mRNAs
upregulated during mitosis and for cell surface functions
676 Cell 151, 671–683, October 26, 2012 ª2012 Elsevier Inc.
(although protein detection was not affected by numbers of
trans-membrane domains; Figure S2F; Table S11). Proteins en-
coded by mitotic mRNAs may only be expressed during a short
cell-cycle window, and thus fall below the detection limit in
unsynchronized cells. Accordingly, of eight cell-cycle-regulated
proteins tested, only two were detectable by fluorescence
microscopy, and they showed expression restricted to specific
phases (Table S12). Small proteins typically had less than or
equal to five MS-compatible peptides and showed lower identi-
fication rates (Figure S2G; Table S11). Taken together, these
data indicate that the proportion of proteins not detected due
to technical limitation (rather than lack of expression) was
<20% of the expressed proteome. Thus, we provide accurate
absolute quantification for most fission yeast proteins, and these
proteins substantially exceed the mRNAs in abundance and
dynamic range.
Coordinated Expression at mRNA and Protein LevelsCopy numbers of mRNAs and corresponding proteins were
highly correlated (Figure 5C). This global relationship between
transcriptome and proteome means that mRNA levels largely
reflect the respective protein levels. Translational properties of
mRNAs, such as ribosome numbers and densities (Lackner
et al., 2007), were also correlated with protein abundance
(R2 �0.1). These data extend previous observations that gene
expression is coordinated at the levels of transcription, mRNA
decay and translation (Lackner et al., 2007) to now also include
protein abundance. However, the ratios between protein and
corresponding mRNA copy numbers spanned over three orders
of magnitude, ranging from 14 to 61,060. This result points to
substantial regulation at the levels of translation and/or protein
turnover. The protein/mRNA ratios were also strongly correlated
with the corresponding protein numbers, but they saturated at
higher protein levels (Figure 5D). This phenomenon suggests
that translation becomes limiting for the most abundant pro-
teins (e.g., owing to saturation of ribosomes on mRNAs), and
that these proteins thus rely on relatively higher mRNA numbers
to boost their abundance. Schwanhausser et al. (2011) have
observed a similar saturation when comparing 5,000 mouse
protein levels to respective translation rates. Notably, the highly
expressed ribosomal proteins formed a distinct group that
showed significantly lower protein/mRNA ratios compared to
genes with similar protein expression (pWilcoxon < 10�9; Fig-
ure 5D). This observation, and related data from Schmidt et al.
(2007), suggests that ribosomal proteins rely more on mRNA
levels than on translation for their high abundance. As ribosomal
proteins act in a complex with rRNAs, the emphasis on transcrip-
tional control might ensure better regulatory coordination with
the nontranslated rRNAs. Other ribonucleoprotein complexes
such as the spliceosome, however, did not show such lower
protein/RNA ratios.
Protein abundance was negatively correlated with protein
length (R2 �0.07), consistent with shorter mRNAs being
more efficiently translated (Lackner et al., 2007). This finding
supports the idea that highly expressed proteins evolved
more streamlined structures due to energetic constraints.
Conversely, no correlation between mRNA abundance and
protein length was evident (R2 �4 3 10�5), suggesting that
any regulatory adaptation occurred at the levels of translation
and/or protein stability, which are energetically more costly
than transcription.
Strikingly, the 20% most abundant proteins accounted for
81.3% of the total protein molecules in proliferating cells, and
this skew was also reflected in the corresponding mRNA
numbers, albeit less pronounced (Figure S2I). This finding
evokes the Pareto principle (‘‘20–80 rule’’) of unequal distribu-
tion in economics and elsewhere, and it highlights that the cell
invests most energy to produce many copies of relatively few
proteins. In addition, the distribution of individual protein
frequencies as a function of their expression rank fitted power-
law distributions, extending a characteristic of mRNA expres-
sion to proteins (Figures S2J and S2K). Taken together, we
conclude that gene regulation is globally coordinated and
streamlined across the expression spectrum.
Protein Abundance in Context of Cellular Landmarksand FunctionsWe compared protein concentrations with cellular ‘‘landmarks’’
for meaningful biological context (Figure 5E). The ribosome is
a large complex composed of single copies of multiple proteins
and rRNAs. Thus, transcriptome and proteome data correctly
calibrated relative to each other should arrive at similar estimates
for total ribosome numbers, allowing cross-validation of our
two independent data sets. Reassuringly, the numbers for
most ribosomal proteins were consistent with the numbers for
different rRNAs (Figure 5E), indicating that there are 1–2 3 105
ribosomes in an average proliferating cell. This number is
comparable to an electron microscopy estimate (�5 3 105 ribo-
somes/cell; Maclean, 1965). For further confirmation, we calcu-
lated the total number of ribosomes associated with mRNAs
by multiplying copy numbers of all individual mRNAs with their
associated ribosome numbers obtained from polysome profiling
(Lackner et al., 2007), resulting in a total of �1.5 3 105 ribo-
somes/cell. Thus, several independent data point to similar
cellular ribosome numbers, corroborating that our quantification
of transcripts and proteins is accurate, both with respect to
absolute numbers and relative to each other. Some ribosomal
proteins showed much lower abundance, however, supporting
the view that they may have nonribosomal functions (Bhavsar
et al., 2010). The median mRNA and protein expression of
single-copy ribosomal proteins was significantly higher than
the median expression of duplicated ribosomal proteins (Fig-
ure S6A); this finding raises the possibility that paralogs
contribute to only part of the ribosome pool, suggesting hetero-
geneous ribosome composition. Proliferating cells contained
approximately four times more ribosomes than mRNAs, illus-
trating the amplification at the level of translation.
The proteasome is a large complex that degrades ubiquiti-
nated proteins. An average cell contained 1–2 3 104 pro-
teasomes, approximately ten times fewer than ribosomes
(Figure 5E). This result highlights that more resources are
invested in protein production than in protein degradation in
proliferating cells that need to continuously produce new pro-
teins to compensate for dilution from cell growth and division.
Proteins of the Pol II transcription complex were present at
7,780 median copies/cell, meaning that cells contain �1 Pol II/
gene (Figure 5E). This low estimate suggests that Pol II could
become limiting, consistent with the finding that Pol II subunit
mutants are haplo-insufficient (Kim et al., 2010), and with
Cell 151, 671–683, October 26, 2012 ª2012 Elsevier Inc. 677
evidence for transcription factories where limiting factors are
concentrated for gene expression (Cook, 2010). One Pol II
subunit, Rpc10, was more highly expressed (20,748 copies/
cell), reflecting that Rpc10 is part of all three RNA polymerase
complexes.
We also analyzed regulatory transcription factors (TFs) that
direct Pol II to specific subsets of genes. The numbers of the
detectable TFs ranged from �100 to >7,000 copies/cell. TFs
controlling meiotic differentiation (Mei4, Atf21, Atf31, Rep1;
Mata et al., 2007) were not detected as proteins and showed
low mRNA abundance in zone 1, whereas the heat shock factor
Hsf1 was the most highly expressed TF (7,244 copies). The large
dynamic range in TF abundance could reflect different mecha-
nisms of transcriptional control, or TF copy numbers might scale
with the numbers of their target genes, although they did not
correlate with the occurrence of known DNAmotifs (Figure S6B).
Proteins of the spliceosome complex were present at 2,675
median copies/cell, similar to the number of intron-containing
genes (Figure 5E). Two splicing proteins, Snu13 and Uap56,
were found at much higher numbers (�1 3 105 and 5 3 104
copies, respectively), probably reflecting their additional roles
in rRNA maturation and mRNA export (Dobbyn and O’Keefe,
2004; Strasser and Hurt, 2001). Thus, the cell produces just
enough spliceosomes to deal with the 2,523 intron-containing
transcriptional units, supporting the view that most splicing
occurs cotranscriptionally in a chromosomal context (de Al-
meida and Carmo-Fonseca, 2012).
Proteins with RNA-recognition motifs (RRM) are an important
class of RNA-binding proteins that control posttranscriptional
gene expression. Intriguingly, RRM protein abundance was
4-fold higher than TF abundance (Figure 5E), evocative of the
4-fold difference between mRNA molecules and protein-coding
genes during G2-phase. Thus, the numbers of RRM proteins
and TFs scale with the numbers of their respective binding
partners. The detected RRM proteins showed large differences
in abundance, ranging from only 175 copies for the methyltrans-
ferase Set1 to 139,690 copies, more than mRNA molecules,
for the uncharacterized Vip1. As for TFs, RRM proteins with
meiotic functions (Mug24, Spo5, Crp79, Mug28, Mde7) where
tightly repressed during proliferation and not detected as
proteins. These findings suggest that some RRM proteins have
transient or specialized roles by targeting few specific tran-
scripts, whereas others have more ubiquitous roles, in line with
genome-wide binding data (Hogan et al., 2008). Accordingly,
the cytoplasmic Pabp, which binds to poly(A) tails of all mRNAs,
showed the second highest expression at 87,000 copies/cell.
This result suggests that approximately two Pabp proteins bind
to average mRNAs, in line with findings that poly(A) tails in yeast
contain�50 residues on average (Lackner et al., 2007) and every
Pabp covers 27 adenine residues (Baer and Kornberg, 1983).
Protein Expression Reflects Cellular FunctionWe also analyzed protein copy numbers with respect to different
functional categories. Discrete patterns of protein expression
distributions were evident, with proteins of different functions
being significantly enriched for distinct abundance ranges (Fig-
ure S7). Thus, proteins of similar functions are often expressed
at similar copy numbers. Three general protein expression
678 Cell 151, 671–683, October 26, 2012 ª2012 Elsevier Inc.
groups were apparent. Lowly expressed proteins (<5,000
copies/cell) were enriched for regulators of biological processes
such as TFs and protein kinases, and for proteins involved
in chromosome structure and DNA repair. An intermediate
group of proteins, expressed at �0.5–1 3 104 copies/cell, often
functioned in RNA metabolism, including splicing, processing
or degradation. Highly expressed proteins (�1 3 104–1.1 3
106 copies/cell) were enriched for functions related to transla-
tion, growth, and metabolism. We conclude that, like mRNAs,
proteins functioning together or in related biological processes
typically share similar expression levels, and these levels reflect
cellular requirements for different tasks or complexes.
Transcriptome Shrinks Globally during QuiescenceTo analyze quantitative RNA and protein changes in a distinct
physiological state, we also acquired data from cells after
24 hr of nitrogen starvation. Upon nitrogen removal, cells stop
growth, divide twice, and arrest as postmitotic, quiescent cells.
These cells remain metabolically active by recycling nitrogen,
become highly resistant to multiple stresses, and survive for
months (Yanagida, 2009).
Quiescent cells are stubby compared to proliferating cells,
showing a median volume reduction of �40%–50% within 12 hr
of nitrogen removal (Figures S4A–S4D). Strikingly, during the
same period, the RNAmass is reduced by�85% to that of prolif-
erating cells (Figure S4E). We measured a total of 89,470 rRNAs/
quiescent cell, representing merely 11.2% of the number in pro-
liferating cells (Figure 6A). The protein-coding transcriptome
showed a somewhat lower reduction, shrinking to 7,419 total
mRNAs (18% of proliferating cells; Figures 6A and 6B). Taking
into account their smaller volumes, quiescent cells contained
�19.6% rRNA and 31.3%mRNA compared to proliferating cells.
The reduction in mRNA copy numbers was global, and remark-
ably coordinated, with abundance in proliferating and quiescent
cells remaining highly correlated (Figure 6C). We conclude that
quiescent cells rely on a substantially smaller transcriptome,
both with respect to RNA abundance and concentration.
Nevertheless, most mRNAs were still expressed within zones
2 or 3 during quiescence (49.8%and 15%, respectively), but with
a median of only 0.69 copies. Thus, although shrinking by 82%
in number, mRNAs retained�72%of the diversity in proliferating
cells. Only 81 mRNAs (1.6% of all) were >2-fold more abundant
in quiescent than in proliferating cells, whereas 4,266 mRNAs
(83.5%) were R2-fold less abundant. Thus, quiescent cells
harbor a diminished but diverse transcriptome, with the majority
of mRNAs being expressed at only �1 copy/cell. It is possible
that low mRNA concentrations represent a more robust expres-
sion during quiescence when cells do not grow and divide, and
mRNAs might be stabilized for long-term endurance (Pluskal
et al., 2011).
Figure 6D compares median mRNA copy numbers for
selected functional categories in proliferating and quiescent
cells. Most categories were substantially downregulated in
quiescence, whereas a few retained similar numbers, including
stress response and sexual differentiation (Figure 6D). Further-
more, three highly expressed categories, all related to protein
translation, were downregulated more than average, yet these
mRNAs remained the most abundant with respect to absolute
Figure 6. Transcriptomes and Proteomes in Proliferating versus Quiescent Cells
(A) Cell volume and rRNA, mRNA, and protein copy numbers in quiescent cells as percentage of corresponding values in proliferating cells.
(B) Distribution of mRNA (left) and protein (right) copies/cell during proliferation (blue) and quiescence (green), with median mRNA and protein abundance during
proliferation and quiescence indicated by horizontal blue and green lines, respectively.
(C) mRNA abundance in quiescent versus proliferating cells. Red and black lines delimit expression zone 2 (0.5–2 mRNA copies/cell) and 2-fold expression
changes, respectively.
(D) Median mRNA abundance of selected functional categories in quiescent versus proliferating cells. Red and black lines as in (C). Red and green dots indicate
lowly and highly repressed categories, respectively (Table S14).
(E) Protein abundance in quiescent versus proliferating cells. Black diagonal lines delimit 2-fold expression changes.
(F) Median protein abundance of selected functional categories in quiescent versus proliferating cells. Black lines as in (E). Red and green dots indicate induced
and repressed categories, respectively (Table S14).
copy numbers (Figure 6D). In conclusion, quiescence is charac-
terized by a global reduction in mRNA numbers, but much less
so in mRNA diversity. mRNAs involved in cell maintenance,
such as adaptation to stress and nutrient limitation, become
relatively more prevalent during quiescence, whereas those
involved in translation become relatively less prevalent, although
they remain highly abundant.
Proteome Does Not Shrink Globally but Is Remodeledduring QuiescenceWe detected a total of 31.2 million protein molecules/quiescent
cell, representing 51.7% of the number measured in proliferating
cells. Adjusting for the decreased volume of quiescent cells,
however, protein numbers were only reduced by �9.5%. Thus,
the proteome largely scaled with volume, and, in stark contrast
to the RNAs, quiescent cells maintained similar protein concen-
trations (Figures 6A and 6B). The median number of protein
copies/quiescent cell was 4,851, which is actually higher than
for proliferating cells. This apparent paradox is explained by
a disproportionate reduction of the 10% most highly expressed
proteins, involved in translation and growth, which account for
87.2% of all proteins lost in quiescent cells (Figures 6B and
6E). We detected 53.2% of all proteins during quiescence,
�13% less than during proliferation. The 897 proteins detected
only in proliferating cells were enriched for mitochondrial trans-
lation and organization (Table S11). This finding suggests that
quiescent cells have decreased oxidative metabolism, consis-
tent with effects on mitochondrial translation and respiration on
chronological lifespan (Pluskal et al., 2011). On the other hand,
the 221 proteins detected only in quiescent cells were enriched
for stress and nitrogen starvation functions.
The proteome was substantially remodeled during quies-
cence, with 47% of all proteins changing their copy numbers
>2-fold (Figure 6E). Figure 6F illustrates this remodeling by
comparing median protein copy numbers for selected functional
categories between proliferating and quiescent cells. Several
Cell 151, 671–683, October 26, 2012 ª2012 Elsevier Inc. 679
Figure 7. Regulatory Dynamics during
Quiescence Entry
(A) Microarray time course to analyze changes in
mRNA levels at 16 time points, before and 30 min
to 7 days after nitrogen removal. Red profiles,
mRNAs induced >1.5-fold within 3 hr after nitrogen
removal; blue profiles, mRNAs repressed
throughout time course. Data are normalized to
0 hr and corrected for total cellular RNA content.
(B) Average expression profiles of stress- and
growth-related genes, and average expression
changes of all genes.
(C) Absolute nCounter measurements of stress-
and growth-related genes, and average profile for
all 49 test genes.
(D) Protein abundance in quiescent versus prolif-
erating cells. Lower right: significance of overlap
between mRNAs induced >1.5-fold within 3 hr
after nitrogen removal (red dots) and proteins
induced >2-fold at 24 hr after nitrogen removal.
cellular maintenance functions, such as stress response,
nitrogen starvation, DNA repair, vacuoles and cell wall, showed
actually increased protein abundance in quiescence (Figure 6F;
Table S14), in stark contrast to the global shrinking observed
for mRNAs (Figure 6D). Categories with decreased protein
abundance were related to translation and growth, similar to
those strongly repressed at the mRNA level (Figure 6F). Notably,
the top 50 most highly expressed proteins during proliferation
were enriched for roles in glucose metabolism and translation,
while during quiescence these proteins were only enriched for
glucose metabolism (Table S11). This finding illustrates that
quiescent cells remain metabolically active, while reducing the
energetic costs of protein synthesis (Shimanuki et al., 2007).
The differences in transcriptome and proteome regulation in
quiescent cells resulted in a lowered correlation between
mRNA and protein copies (R2 = 0.36) compared to proliferating
cells (R2 = 0.55).
In conclusion, quiescent cells upregulate proteins implicated
in a dormant lifestyle, while maintaining an abundant, yet
strongly reduced, translational machinery. Together with the
drastic reduction in overall mRNA abundance, this finding
highlights the change in cellular physiology from a growth
program for proliferation to a maintenance program for stress
680 Cell 151, 671–683, October 26, 2012 ª2012 Elsevier Inc.
protection and long-term endurance.
These two fundamental programs are im-
plemented by balancing the expression
of stress- versus growth-related genes,
regulated by antagonistic signaling path-
ways such as the stress-activated protein
kinase (SAPK) and target of rapamycin
(TOR) (Lopez-Maury et al., 2008).
Early mRNA Burst Sustains HighProtein Numbers duringQuiescenceWe showed that 24 hr after nitrogen
removal quiescent cells reached a state
of a globally diminished transcriptome and a remodeled pro-
teome. How do these cells manage to upregulate numerous
proteins while downregulating most of the corresponding
mRNAs (Figure 6)? We pursued this question by analyzing
dynamic changes in mRNA levels at high temporal resolution.
This time course experiment revealed that within 12 hr of
nitrogen removal most mRNAs decreased whereas others
transiently increased, followed by largely constant mRNA levels
from 12–186 hr (Figure 7A). Although many stress-related genes
were induced within 2 hr of nitrogen removal before becom-
ing repressed, growth-related genes became immediately
repressed (Figure 7B). We also measured absolute mRNA abun-
dance for 49 genes from the same cell samples that reiterated
the global data, excluding a normalization artifact (Figure 7C).
Note that the average expression of all genes, and the absolute
expression of the 49 test genes, decreased during the time
course, with the stress-related genes showing a lower decrease
and the growth-related mRNAs a higher decrease relative to all
genes (Figures 7B and 7C). This pattern is also reflected in
relative expression changes from microarray data (Mata and
Bahler, 2006; Shimanuki et al., 2007). The absolute data pre-
sented here, however, expands and refines our understanding
of this gene expression program, revealing that the upregulation
of stress-related genes is only transient, followed by a global
repression of most genes.
Entry into quiescence thus consists of two phases: (1) a rapid
adaptation where selected genes are induced, and (2) a global,
but differential, repression of most genes. The burst in stress-
related mRNAs could contribute to the proteome reshuffling
observed at 24 hr. Indeed, a significant number of proteins
with increased levels during quiescence corresponded to the
transiently-induced mRNAs (Figure 7D). Thus, the early mRNA
burst leads to a sustained increase of selected proteins, long
after the corresponding mRNAs have decreased again. This
mode of regulation depends on longer half-lives for proteins
than for mRNAs, and it is plausible that proteins become further
stabilized during quiescence. We conclude that cells, upon
nitrogen removal, immediately repress the growth-related
mRNAs while transiently inducing stress-related mRNAs, which
in turn help to adjust the proteome for extended quiescence.
ConclusionsWe comprehensively quantified the average numbers of RNAs
and proteins in two fundamental cellular states of a eukaryotic
model system. Besides providing a lasting resource for follow-
up-studies, our data provide unique insight into cell regulation
and function. Although mRNA and protein levels are well corre-
lated overall, more strongly in proliferating than in quiescent
cells, different mRNAs are �10 to 60,000-fold less abundant
than the corresponding proteins. This finding highlights the
substantial amplification and regulation occurring during trans-
lation and protein turnover. Given that most RNAs are ex-
pressed at single-digit copy numbers, they are much more
susceptible to stochastic events than proteins expressed at
thousands of copies. Distinct expression zones for mRNAs
and proteins reflect functional demands. Most mRNAs are ex-
pressed at �1–10 copies/cell, whereas mRNAs present at
well below 1 copy/cell are enriched for tightly repressed differ-
entiation and regulatory genes that typically do not produce
detectable proteins. This finding contrasts with data from
bacteria where productive protein expression is achieved with
such low mRNA concentrations (Taniguchi et al., 2010). Ulti-
mately, population average measurements will need to be inte-
grated with single-cell data to understand more complex
cellular distributions of RNAs and proteins (Hebenstreit et al.,
2011).
lncRNAs are generally present at well below 1 copy/cell,
although �200 lncRNAs, including �40 nonpolyadenylated
RNAs, are more robustly expressed at �1–200 copies/cell and
are thus prime candidates for functional analyses. However,
the abundance of these lncRNAs is still much lower than for
most proteins, suggesting functions with different biochemical
characteristics. For instance, �1 lncRNA copy/cell could be
sufficient for roles in cis, where the RNA acts where it is tran-
scribed, whereas higher expression levels could suggest roles
in trans.
The transcriptome is larger in proliferating than in quiescent
cells, reflecting the higher need for transcription during growth
and division, and suggesting the existence of a global regulatory
mechanism coordinating overall RNA abundance. In contrast,
the proteome size is similar in proliferating and quiescent cells,
after adjusting for differences in cell volume. However, the rela-
tive levels of numerous proteins show striking antagonistic
changes in proliferating and quiescent cells, adapted for cellular
growth or maintenance, respectively. Proteome remodeling
during quiescence is enabled by a transient burst of stress-
related mRNAs that leads to sustained high levels of the corre-
sponding proteins. Protein concentrations are optimized to
avoid molecular crowding for biochemical reactions (Dill et al.,
2011). Constant protein concentrations during growth imply
that protein numbers increase with cell volume, as do rRNA
and ribosome numbers (Maclean, 1965) as well as mRNA
numbers (Zhurinsky et al., 2010). Thus, the absolute cellular
numbers of mRNA and protein molecules are not fixed by the
genome but are globally tuned to cell volume and physiology.
Genome-wide data on RNA and protein quantities are therefore
vital to decipher the complex relationships linking genome
regulation with cell physiology and growth, and to understand
how different cells with identical genomes achieve the enor-
mous diversity of functions. The findings reported here highlight
elementary features of transcriptome and proteome regulation
and provide a valuable platform to support future studies and
quantitative biology.
EXPERIMENTAL PROCEDURES
Full methods are available in Extended Experimental Procedures.
Cell Cultures
Wild-type 972 h� fission yeast cells were grown in Edinburgh minimal medium
(EMM) at 32�C to mid-log phase; for quiescence experiments, such cells
were shifted to EMM without nitrogen at 32�C and harvested at different
times after nitrogen removal. Several cell pellets from the same cultures
were frozen and used for RNA-seq, nCounter, and proteomics.
Quantitative Transcriptomics
RNA was extracted using the hot-phenol technique. Strand-specific RNA-seq
libraries were prepared from total or poly(A)+ RNA using an early version of
the Illumina TruSeq Small RNA Sample Prep Kit. Sequencing scores were
calculated as number of reads/kilobase. Scores derived from total RNA
libraries were calibrated using absolute data acquired for 49 mRNAs, in whole
cell extracts, on a nCounter instrument (NanoString), with external controls
spiked in known quantities.
Quantitative Proteomics
Extracted proteins were enzymatically digested using trypsin, the peptides
were separated into 12 fractions using an OFF-GEL Fractionator (Agilent),
and analyzed on an Orbitrap Velos LC-MS platform (Thermo Scientific).
Peptides were quantified and identified using the Progenesis LC-MS
(Nonlinear Dynamics) andMascot software, respectively. Absolute abundance
for 39 proteins was determined using spiked-in heavy reference peptides
to translate the summed MS-intensities of all peptides to copies/cell for all
identified proteins.
Modeling of Cell-Cycle-Regulated mRNA Abundance
Periodic mRNA abundance was modeled for different cell-cycle phases
using (1) average mRNA copies/cell in asynchronous cultures, (2) fraction of
cell cycle with mRNA peak expression, and (3) amplitude of periodic mRNA
regulation.
Quiescence Entry Time Course
RNA was extracted using the hot-phenol technique. Labeled cDNA of each
sample was hybridized against a pool of all samples on a custom-designed
Cell 151, 671–683, October 26, 2012 ª2012 Elsevier Inc. 681
Agilent microarray. Absolute data for 49 test genes were acquired from the
same cell samples using nCounter as described above.
ACCESSION NUMBERS
The ArrayExpress accession numbers for the RNA-seq and microarray data
reported in this paper are MTAB-1154, and E-TABM-1075, respectively. The
LC-MS data are available from ProteomeCommons.org Tranche using
these hashes:
Proliferating cells OGE-fractions:
jsVcj9T6yAfpS80nYRueuobP0iOdpQzk/IgyYCYf+EZpj6fAE7dx9JoKvkhCfmQ
J5d5NuGLfriFt YEKJcB4rw+egFFcAAAAAAAACAw==
Quiescent cells OGE-fractions:
1gNlqHIUVaHU8S8yqpR8tvKvq6m0jySALyTlBxte1YvIo/N8IPXlMsgRKu2ZTg
sdPHPG0Z0jrrWtiDJ1CEXSb/fQY0kAAAAAAAACHg==
Unfractionated samples:
rSK2uX3cA5h/YrSal0U3y6Y0VneHG7p3J6fdXWAeLYC6HbnIejsgoKTgUNZL
We2Q+GRNQcc+xogxlG723ayC88ONM1cAAAAAAAAClg==
SUPPLEMENTAL INFORMATION
Supplemental information includes Extended Experimental Procedures, seven
figures, and 18 tables and can be found with this article online at http://dx.doi.
org/10.1016/j.cell.2012.09.019.
ACKNOWLEDGMENTS
We thank F. Bachand, D. Bitton, L. Foukas, A. Lock, D. Lovell, J. Mata, and V.
Wood for comments on the manuscript, P. Gardner for advice on lncRNA
sequences, and P. Oliveri for help with nCounter. This research was funded
by PhenOxiGEn (an EU FP7 research project), a BBSRC Research Grant
BB/I012451/1, and a Wellcome Trust Senior Investigator Award to J.B., and
by SystemsX.ch and an ERC Advanced Grant to R.A.
Received: February 10, 2012
Revised: June 11, 2012
Accepted: July 26, 2012
Published: October 25, 2012
REFERENCES
Atkinson, S.R., Marguerat, S., and Bahler, J. (2012). Exploring long non-coding
RNAs through sequencing. Semin. Cell Dev. Biol. 23, 200–205.
Baer, B.W., and Kornberg, R.D. (1983). The protein responsible for the
repeating structure of cytoplasmic poly(A)-ribonucleoprotein. J. Cell Biol. 96,
717–721.
Beck, M., Claassen, M., and Aebersold, R. (2011). Comprehensive proteo-
mics. Curr. Opin. Biotechnol. 22, 3–8.
Bhavsar, R.B., Makley, L.N., and Tsonis, P.A. (2010). The other lives of
ribosomal proteins. Hum. Genomics 4, 327–344.
Chen, D., Toone, W.M., Mata, J., Lyne, R., Burns, G., Kivinen, K., Brazma, A.,
Jones, N., and Bahler, J. (2003). Global transcriptional responses of fission
yeast to environmental stress. Mol. Biol. Cell 14, 214–229.
Coller, H.A. (2011). Cell biology. The essence of quiescence. Science 334,
1074–1075.
Cook, P.R. (2010). A model for all genomes: the role of transcription factories.
J. Mol. Biol. 395, 1–10.
Cox, J., and Mann, M. (2011). Quantitative, high-resolution proteomics for
data-driven systems biology. Annu. Rev. Biochem. 80, 273–299.
de Almeida, S.F., and Carmo-Fonseca, M. (2012). Design principles of inter-
connections between chromatin and pre-mRNA splicing. Trends Biochem.
Sci. 37, 248–253.
682 Cell 151, 671–683, October 26, 2012 ª2012 Elsevier Inc.
Dill, K.A., Ghosh, K., and Schmit, J.D. (2011). Physical limits of cells and
proteomes. Proc. Natl. Acad. Sci. USA 108, 17876–17882.
Dobbyn, H.C., and O’Keefe, R.T. (2004). Analysis of Snu13p mutations
reveals differential interactions with the U4 snRNA and U3 snoRNA. RNA 10,
308–320.
Gardner, P.P., Daub, J., Tate, J., Moore, B.L., Osuch, I.H., Griffiths-Jones, S.,
Finn, R.D., Nawrocki, E.P., Kolbe, D.L., Eddy, S.R., and Bateman, A. (2011).
Rfam: Wikipedia, clans and the ‘‘decimal’’ release. Nucleic Acids Res.
39(Database issue), D141–D145.
Grewal, S.I. (2010). RNAi-dependent formation of heterochromatin and its
diverse functions. Curr. Opin. Genet. Dev. 20, 134–141.
Harigaya, Y., and Yamamoto, M. (2007). Molecular mechanisms underlying
the mitosis-meiosis decision. Chromosome Res. 15, 523–537.
Hebenstreit, D., Fang, M., Gu, M., Charoensawan, V., van Oudenaarden, A.,
and Teichmann, S.A. (2011). RNA sequencing reveals two major classes of
gene expression levels in metazoan cells. Mol. Syst. Biol. 7, 497.
Hogan, D.J., Riordan, D.P., Gerber, A.P., Herschlag, D., and Brown, P.O.
(2008). Diverse RNA-binding proteins interact with functionally related sets
of RNAs, suggesting an extensive regulatory system. PLoS Biol. 6, e255.
Holstege, F.C., Jennings, E.G., Wyrick, J.J., Lee, T.I., Hengartner, C.J., Green,
M.R., Golub, T.R., Lander, E.S., and Young, R.A. (1998). Dissecting the
regulatory circuitry of a eukaryotic genome. Cell 95, 717–728.
Kim, D.U., Hayles, J., Kim, D.,Wood, V., Park, H.O.,Won,M., Yoo, H.S., Duhig,
T., Nam, M., Palmer, G., et al. (2010). Analysis of a genome-wide set of gene
deletions in the fission yeast Schizosaccharomyces pombe. Nat. Biotechnol.
28, 617–623.
Lackner, D.H., Beilharz, T.H., Marguerat, S., Mata, J., Watt, S., Schubert, F.,
Preiss, T., and Bahler, J. (2007). A network of multiple regulatory layers shapes
gene expression in fission yeast. Mol. Cell 26, 145–155.
Larson, D.R., Singer, R.H., and Zenklusen, D. (2009). A single molecule view of
gene expression. Trends Cell Biol. 19, 630–637.
Lemay, J.F., D’Amours, A., Lemieux, C., Lackner, D.H., St-Sauveur, V.G.,
Bahler, J., and Bachand, F. (2010). The nuclear poly(A)-binding protein inter-
acts with the exosome to promote synthesis of noncoding small nucleolar
RNAs. Mol. Cell 37, 34–45.
Li, L., and Clevers, H. (2010). Coexistence of quiescent and active adult
stem cells in mammals. Science 327, 542–545.
Lopez-Maury, L., Marguerat, S., and Bahler, J. (2008). Tuning gene expression
to changing environments: from rapid responses to evolutionary adaptation.
Nat. Rev. Genet. 9, 583–593.
Lovell, D., Muller, W., Taylor, J., Zwart, A., and Helliwell, C. (2011). Proportion,
percentages, PPM: do the molecular biosciences treat compositional data
right? In Compositional Data Analysis: Theory and Applications, V. Pawlow-
sky-Glahn, A. Buccianti, D. Lovell, W. Muller, J. Taylor, A. Zwart, and C. Helli-
well, eds. (New York: John Wiley & Sons), pp. 193–206.
Maclean, N. (1965). Ribosome numbers in a fission yeast. Nature 207,
322–323.
Maier, T., Schmidt, A., Guell, M., Kuhner, S., Gavin, A.C., Aebersold, R., and
Serrano, L. (2011). Quantification of mRNA and protein and integration with
protein turnover in a bacterium. Mol. Syst. Biol. 7, 511.
Marguerat, S., Jensen, T.S., de Lichtenberg, U., Wilhelm, B.T., Jensen, L.J.,
andBahler, J. (2006). Themore themerrier: comparative analysis ofmicroarray
studies on cell cycle-regulated genes in fission yeast. Yeast 23, 261–277.
Mata, J., and Bahler, J. (2006). Global roles of Ste11p, cell type, and phero-
mone in the control of gene expression during early sexual differentiation in
fission yeast. Proc. Natl. Acad. Sci. USA 103, 15517–15522.
Mata, J., Lyne, R., Burns, G., andBahler, J. (2002). The transcriptional program
of meiosis and sporulation in fission yeast. Nat. Genet. 32, 143–147.
Mata, J., Wilbrey, A., and Bahler, J. (2007). Transcriptional regulatory network
for sexual differentiation in fission yeast. Genome Biol. 8, R217.
McPheeters, D.S., Cremona, N., Sunder, S., Chen, H.M., Averbeck, N., Leath-
erwood, J., and Wise, J.A. (2009). A complex gene regulatory mechanism that
operates at the nexus of multiple RNA processing decisions. Nat. Struct. Mol.
Biol. 16, 255–264.
Miura, F., Kawaguchi, N., Yoshida, M., Uematsu, C., Kito, K., Sakaki, Y., and
Ito, T. (2008). Absolute quantification of the budding yeast transcriptome by
means of competitive PCR between genomic and complementary DNAs.
BMC Genomics 9, 574.
Nagaraj, N., Wisniewski, J.R., Geiger, T., Cox, J., Kircher, M., Kelso, J., Paabo,
S., and Mann, M. (2011). Deep proteome and transcriptome mapping of
a human cancer cell line. Mol. Syst. Biol. 7, 548.
Ng, S.S., Anderson, M., White, S., and McInerny, C.J. (2001). mik1(+) G1-S
transcription regulates mitotic entry in fission yeast. FEBS Lett. 503, 131–134.
Ozsolak, F., and Milos, P.M. (2011). RNA sequencing: advances, challenges
and opportunities. Nat. Rev. Genet. 12, 87–98.
Pancaldi, V., Schubert, F., and Bahler, J. (2010). Meta-analysis of genome
regulation and expression variability across hundreds of environmental and
genetic perturbations in fission yeast. Mol. Biosyst. 6, 543–552.
Pluskal, T., Hayashi, T., Saitoh, S., Fujisawa, A., and Yanagida, M. (2011).
Specific biomarkers for stochastic division patterns and starvation-induced
quiescence under limited glucose levels in fission yeast. FEBS J. 278, 1299–
1315.
Rhind, N., Chen, Z., Yassour, M., Thompson, D.A., Haas, B.J., Habib, N.,
Wapinski, I., Roy, S., Lin, M.F., Heiman, D.I., et al. (2011). Comparative
functional genomics of the fission yeasts. Science 332, 930–936.
Rustici, G., Mata, J., Kivinen, K., Lio, P., Penkett, C.J., Burns, G., Hayles, J.,
Brazma, A., Nurse, P., and Bahler, J. (2004). Periodic gene expression
program of the fission yeast cell cycle. Nat. Genet. 36, 809–817.
Schmidt, A., Beck, M., Malmstrom, J., Lam, H., Claassen, M., Campbell, D.,
and Aebersold, R. (2011). Absolute quantification of microbial proteomes at
different states by directed mass spectrometry. Mol. Syst. Biol. 7, 510.
Schmidt, M.W., Houseman, A., Ivanov, A.R., and Wolf, D.A. (2007). Compar-
ative proteomic and transcriptomic profiling of the fission yeast Schizosac-
charomyces pombe. Mol. Syst. Biol. 3, 79.
Schwanhausser, B., Busse, D., Li, N., Dittmar, G., Schuchhardt, J., Wolf, J.,
Chen, W., and Selbach, M. (2011). Global quantification of mammalian gene
expression control. Nature 473, 337–342.
Shimanuki, M., Chung, S.Y., Chikashige, Y., Kawasaki, Y., Uehara, L., Tsut-
sumi, C., Hatanaka, M., Hiraoka, Y., Nagao, K., and Yanagida, M. (2007).
Two-step, extensive alterations in the transcriptome from G0 arrest to cell
division in Schizosaccharomyces pombe. Genes Cells 12, 677–692.
Strasser, K., and Hurt, E. (2001). Splicing factor Sub2p is required for nuclear
mRNA export through its interaction with Yra1p. Nature 413, 648–652.
Taniguchi, Y., Choi, P.J., Li, G.W., Chen, H., Babu, M., Hearn, J., Emili, A., and
Xie, X.S. (2010). Quantifying E. coli proteome and transcriptome with single-
molecule sensitivity in single cells. Science 329, 533–538.
Vogel, C., and Marcotte, E.M. (2012). Insights into the regulation of protein
abundance from proteomic and transcriptomic analyses. Nat. Rev. Genet.
13, 227–232.
Wilhelm, B.T., Marguerat, S., Watt, S., Schubert, F., Wood, V., Goodhead, I.,
Penkett, C.J., Rogers, J., and Bahler, J. (2008). Dynamic repertoire of a
eukaryotic transcriptome surveyed at single-nucleotide resolution. Nature
453, 1239–1243.
Wood, V., Harris, M.A., McDowall, M.D., Rutherford, K., Vaughan, B.W.,
Staines, D.M., Aslett, M., Lock, A., Bahler, J., Kersey, P.J., and Oliver, S.G.
(2012). PomBase: a comprehensive online resource for fission yeast. Nucleic
Acids Res. 40(Database issue), D695–D699.
Wu, J.Q., and Pollard, T.D. (2005). Counting cytokinesis proteins globally and
locally in fission yeast. Science 310, 310–314.
Yanagida, M. (2009). Cellular quiescence: are controlling genes conserved?
Trends Cell Biol. 19, 705–715.
Zenklusen, D., Larson, D.R., and Singer, R.H. (2008). Single-RNA counting
reveals alternative modes of gene expression in yeast. Nat. Struct. Mol. Biol.
15, 1263–1271.
Zhurinsky, J., Leonhard, K., Watt, S., Marguerat, S., Bahler, J., and Nurse, P.
(2010). A coordinated global control over cellular transcription. Curr. Biol. 20,
2010–2015.
Zofall, M., Yamanaka, S., Reyes-Turcu, F.E., Zhang, K., Rubin, C., and Grewal,
S.I. (2012). RNA elimination machinery targeting meiotic mRNAs promotes
facultative heterochromatin formation. Science 335, 96–100.
Cell 151, 671–683, October 26, 2012 ª2012 Elsevier Inc. 683
